A Novel Dynamic Spectrum Access Strategy Applied to Cognitive Radio Network

This paper proposes a dynamic spectrum access (DSA) strategy in cognitive radio (CR) network. This strategy combines fuzzy inference and reinforcement learning theory to build knowledge on network status from the current state and past experience. Multiple assessment variables on environment performance and preference information are introduced to regulate decision-making behavior. The proposed aims at making optimal spectrum access action and obtaining long-term rewards on desired network-wide objectives. We adopt distribute DSA architecture in which cognitive radio network access points (AP) executive control strategy and provide CR network access for unlicensed users. Numerical results show that our algorithm can dramatically promote efficient use of spectrum resource compared to other spectrum access methods, and gain good rewards on the desirable objectives.